Sensitivity study of SPARC H-mode scenarios performance with reduced transport models

The prediction of transport and confinement is crucial to quantify the fusion performance of existing and future machines. It has traditionally been based on empirical scaling laws but physics-based modeling offers the possibility of more accurate extrapolations from current experiments. In the present work, an extensive database of SPARC predictions has been obtained in the ASTRA framework, using the quasilinear transport model TGLF SAT2, including electromagnetic effects, and a neural network (NN) fitted on EPED results for the pedestal stability. The database has been obtained starting from two H-mode reference discharges at full 12.2T (SPARC PRD) and reduced 8T field. Some input parameters (W and DT concentration, edge Ti/Te and p_pedestal/p_EPED) are typically uncertain and hard to control, therefore they have been randomly assigned to assess their effect on the performance. Additionally, a scan of auxiliary input power (ion cyclotron heating) and density at top of pedestal has been performed to verify if highly radiative plasmas are above the LH power threshold. Despite a big scatter of the performance parameters, the database has shown promising results: the PRD database shows often burning plasma conditions (Q > 5) and always have Q > 2, which is the overall goal of SPARC; the 8T H-mode database often meets Q > 1 conditions, that, coupled with its capability to be run safely far from the engineering limits, makes it a potential candidate as an early operation scenario to reach breakeven in SPARC.